fix invalidates media url's

This commit is contained in:
Maksim Pechnikov 2020-06-14 21:02:57 +03:00
parent f9dcf15ecb
commit 2e8a236cef
16 changed files with 346 additions and 114 deletions

View file

@ -406,6 +406,13 @@
],
whitelist: []
config :pleroma, Pleroma.Web.MediaProxy.Invalidation.Http,
method: :purge,
headers: [],
options: []
config :pleroma, Pleroma.Web.MediaProxy.Invalidation.Script, script_path: nil
config :pleroma, :chat, enabled: true
config :phoenix, :format_encoders, json: Jason

View file

@ -1637,6 +1637,31 @@
"The base URL to access a user-uploaded file. Useful when you want to proxy the media files via another host/CDN fronts.",
suggestions: ["https://example.com"]
},
%{
key: :invalidation,
type: :keyword,
descpiption: "",
suggestions: [
enabled: true,
provider: Pleroma.Web.MediaProxy.Invalidation.Script
],
children: [
%{
key: :enabled,
type: :boolean,
description: "Enables invalidate media cache"
},
%{
key: :provider,
type: :module,
description: "Module which will be used to cache purge.",
suggestions: [
Pleroma.Web.MediaProxy.Invalidation.Script,
Pleroma.Web.MediaProxy.Invalidation.Http
]
}
]
},
%{
key: :proxy_opts,
type: :keyword,
@ -1709,6 +1734,45 @@
}
]
},
%{
group: :pleroma,
key: Pleroma.Web.MediaProxy.Invalidation.Http,
type: :group,
description: "HTTP invalidate settings",
children: [
%{
key: :method,
type: :atom,
description: "HTTP method of request. Default: :purge"
},
%{
key: :headers,
type: {:list, :tuple},
description: "HTTP headers of request.",
suggestions: [{"x-refresh", 1}]
},
%{
key: :options,
type: :keyword,
description: "Request options.",
suggestions: [params: %{ts: "xxx"}]
}
]
},
%{
group: :pleroma,
key: Pleroma.Web.MediaProxy.Invalidation.Script,
type: :group,
description: "Script invalidate settings",
children: [
%{
key: :script_path,
type: :string,
description: "Path to shell script. Which will run purge cache.",
suggestions: ["./installation/nginx-cache-purge.sh.example"]
}
]
},
%{
group: :pleroma,
key: :gopher,

View file

@ -262,7 +262,7 @@ This section describe PWA manifest instance-specific values. Currently this opti
#### Pleroma.Web.MediaProxy.Invalidation.Script
This strategy allow perform external bash script to purge cache.
This strategy allow perform external shell script to purge cache.
Urls of attachments pass to script as arguments.
* `script_path`: path to external script.
@ -278,8 +278,8 @@ config :pleroma, Pleroma.Web.MediaProxy.Invalidation.Script,
This strategy allow perform custom http request to purge cache.
* `method`: http method. default is `purge`
* `headers`: http headers. default is empty
* `options`: request options. default is empty
* `headers`: http headers.
* `options`: request options.
Example:
```elixir

View file

@ -13,7 +13,7 @@ CACHE_DIRECTORY="/tmp/pleroma-media-cache"
## $3 - (optional) the number of parallel processes to run for grep.
get_cache_files() {
local max_parallel=${3-16}
find $2 -maxdepth 2 -type d | xargs -P $max_parallel -n 1 grep -E Rl "^KEY:.*$1" | sort -u
find $2 -maxdepth 2 -type d | xargs -P $max_parallel -n 1 grep -E -Rl "^KEY:.*$1" | sort -u
}
## Removes an item from the given cache zone.
@ -37,4 +37,4 @@ purge() {
}
purge $1
purge $@

View file

@ -148,7 +148,8 @@ defp cachex_children do
build_cachex("idempotency", expiration: idempotency_expiration(), limit: 2500),
build_cachex("web_resp", limit: 2500),
build_cachex("emoji_packs", expiration: emoji_packs_expiration(), limit: 10),
build_cachex("failed_proxy_url", limit: 2500)
build_cachex("failed_proxy_url", limit: 2500),
build_cachex("deleted_urls", default_ttl: :timer.hours(24 * 30), limit: 5_000)
]
end

View file

@ -10,6 +10,8 @@ defmodule Pleroma.Plugs.UploadedMedia do
import Pleroma.Web.Gettext
require Logger
alias Pleroma.Web.MediaProxy
@behaviour Plug
# no slashes
@path "media"
@ -35,8 +37,7 @@ def call(%{request_path: <<"/", @path, "/", file::binary>>} = conn, opts) do
%{query_params: %{"name" => name}} = conn ->
name = String.replace(name, "\"", "\\\"")
conn
|> put_resp_header("content-disposition", "filename=\"#{name}\"")
put_resp_header(conn, "content-disposition", "filename=\"#{name}\"")
conn ->
conn
@ -47,7 +48,8 @@ def call(%{request_path: <<"/", @path, "/", file::binary>>} = conn, opts) do
with uploader <- Keyword.fetch!(config, :uploader),
proxy_remote = Keyword.get(config, :proxy_remote, false),
{:ok, get_method} <- uploader.get_file(file) do
{:ok, get_method} <- uploader.get_file(file),
false <- media_is_deleted(conn, get_method) do
get_media(conn, get_method, proxy_remote, opts)
else
_ ->
@ -59,6 +61,14 @@ def call(%{request_path: <<"/", @path, "/", file::binary>>} = conn, opts) do
def call(conn, _opts), do: conn
defp media_is_deleted(%{request_path: path} = _conn, {:static_dir, _}) do
MediaProxy.in_deleted_urls(Pleroma.Web.base_url() <> path)
end
defp media_is_deleted(_, {:url, url}), do: MediaProxy.in_deleted_urls(url)
defp media_is_deleted(_, _), do: false
defp get_media(conn, {:static_dir, directory}, _, opts) do
static_opts =
Map.get(opts, :static_plug_opts)

View file

@ -5,22 +5,33 @@
defmodule Pleroma.Web.MediaProxy.Invalidation do
@moduledoc false
@callback purge(list(String.t()), map()) :: {:ok, String.t()} | {:error, String.t()}
@callback purge(list(String.t()), Keyword.t()) :: {:ok, list(String.t())} | {:error, String.t()}
alias Pleroma.Config
alias Pleroma.Web.MediaProxy
@spec purge(list(String.t())) :: {:ok, String.t()} | {:error, String.t()}
@spec enabled?() :: boolean()
def enabled?, do: Config.get([:media_proxy, :invalidation, :enabled])
@spec purge(list(String.t()) | String.t()) :: {:ok, list(String.t())} | {:error, String.t()}
def purge(urls) do
[:media_proxy, :invalidation, :enabled]
|> Config.get()
|> do_purge(urls)
prepared_urls = prepare_urls(urls)
if enabled?() do
do_purge(prepared_urls)
else
{:ok, prepared_urls}
end
end
defp do_purge(true, urls) do
defp do_purge(urls) do
provider = Config.get([:media_proxy, :invalidation, :provider])
options = Config.get(provider)
provider.purge(urls, options)
provider.purge(urls, Config.get(provider))
end
defp do_purge(_, _), do: :ok
def prepare_urls(urls) do
urls
|> List.wrap()
|> Enum.map(&MediaProxy.url(&1))
end
end

View file

@ -10,9 +10,9 @@ defmodule Pleroma.Web.MediaProxy.Invalidation.Http do
@impl Pleroma.Web.MediaProxy.Invalidation
def purge(urls, opts) do
method = Map.get(opts, :method, :purge)
headers = Map.get(opts, :headers, [])
options = Map.get(opts, :options, [])
method = Keyword.get(opts, :method, :purge)
headers = Keyword.get(opts, :headers, [])
options = Keyword.get(opts, :options, [])
Logger.debug("Running cache purge: #{inspect(urls)}")
@ -22,7 +22,7 @@ def purge(urls, opts) do
end
end)
{:ok, "success"}
{:ok, urls}
end
defp do_purge(method, url, headers, options) do

View file

@ -10,32 +10,34 @@ defmodule Pleroma.Web.MediaProxy.Invalidation.Script do
require Logger
@impl Pleroma.Web.MediaProxy.Invalidation
def purge(urls, %{script_path: script_path} = _options) do
def purge(urls, opts) do
args =
urls
|> List.wrap()
|> Enum.uniq()
|> Enum.join(" ")
path = Path.expand(script_path)
Logger.debug("Running cache purge: #{inspect(urls)}, #{path}")
case do_purge(path, [args]) do
{result, exit_status} when exit_status > 0 ->
Logger.error("Error while cache purge: #{inspect(result)}")
{:error, inspect(result)}
_ ->
{:ok, "success"}
end
opts
|> Keyword.get(:script_path, nil)
|> do_purge([args])
|> handle_result(urls)
end
def purge(_, _), do: {:error, "not found script path"}
defp do_purge(path, args) do
defp do_purge(script_path, args) when is_binary(script_path) do
path = Path.expand(script_path)
Logger.debug("Running cache purge: #{inspect(args)}, #{inspect(path)}")
System.cmd(path, args)
rescue
error -> {inspect(error), 1}
error -> error
end
defp do_purge(_, _), do: {:error, "not found script path"}
defp handle_result({_result, 0}, urls), do: {:ok, urls}
defp handle_result({:error, error}, urls), do: handle_result(error, urls)
defp handle_result(error, _) do
Logger.error("Error while cache purge: #{inspect(error)}")
{:error, inspect(error)}
end
end

View file

@ -6,20 +6,53 @@ defmodule Pleroma.Web.MediaProxy do
alias Pleroma.Config
alias Pleroma.Upload
alias Pleroma.Web
alias Pleroma.Web.MediaProxy.Invalidation
@base64_opts [padding: false]
@spec in_deleted_urls(String.t()) :: boolean()
def in_deleted_urls(url), do: elem(Cachex.exists?(:deleted_urls_cache, url(url)), 1)
def remove_from_deleted_urls(urls) when is_list(urls) do
Cachex.execute!(:deleted_urls_cache, fn cache ->
Enum.each(Invalidation.prepare_urls(urls), &Cachex.del(cache, &1))
end)
end
def remove_from_deleted_urls(url) when is_binary(url) do
Cachex.del(:deleted_urls_cache, url(url))
end
def put_in_deleted_urls(urls) when is_list(urls) do
Cachex.execute!(:deleted_urls_cache, fn cache ->
Enum.each(Invalidation.prepare_urls(urls), &Cachex.put(cache, &1, true))
end)
end
def put_in_deleted_urls(url) when is_binary(url) do
Cachex.put(:deleted_urls_cache, url(url), true)
end
def url(url) when is_nil(url) or url == "", do: nil
def url("/" <> _ = url), do: url
def url(url) do
if disabled?() or local?(url) or whitelisted?(url) do
if disabled?() or not is_url_proxiable?(url) do
url
else
encode_url(url)
end
end
@spec is_url_proxiable?(String.t()) :: boolean()
def is_url_proxiable?(url) do
if local?(url) or whitelisted?(url) do
false
else
true
end
end
defp disabled?, do: !Config.get([:media_proxy, :enabled], false)
defp local?(url), do: String.starts_with?(url, Pleroma.Web.base_url())

View file

@ -14,10 +14,11 @@ def remote(conn, %{"sig" => sig64, "url" => url64} = params) do
with config <- Pleroma.Config.get([:media_proxy], []),
true <- Keyword.get(config, :enabled, false),
{:ok, url} <- MediaProxy.decode_url(sig64, url64),
{_, false} <- {:in_deleted_urls, MediaProxy.in_deleted_urls(url)},
:ok <- filename_matches(params, conn.request_path, url) do
ReverseProxy.call(conn, url, Keyword.get(config, :proxy_opts, @default_proxy_opts))
else
false ->
error when error in [false, {:in_deleted_urls, true}] ->
send_resp(conn, 404, Plug.Conn.Status.reason_phrase(404))
{:error, :invalid_signature} ->

View file

@ -23,8 +23,25 @@ def perform(
Enum.map(attachment["url"], & &1["href"])
end)
names = Enum.map(attachments, & &1["name"])
# find all objects for copies of the attachments, name and actor doesn't matter here
hrefs
|> fetch_objects
|> prepare_objects(actor, Enum.map(attachments, & &1["name"]))
|> Enum.reduce({[], []}, fn {href, %{id: id, count: count}}, {ids, hrefs} ->
with 1 <- count do
{ids ++ [id], hrefs ++ [href]}
else
_ -> {ids ++ [id], hrefs}
end
end)
|> do_clean
{:ok, :success}
end
def perform(%{"op" => "cleanup_attachments", "object" => _object}, _job), do: {:ok, :skip}
defp do_clean({object_ids, attachment_urls}) do
uploader = Pleroma.Config.get([Pleroma.Upload, :uploader])
prefix =
@ -39,68 +56,60 @@ def perform(
"/"
)
# find all objects for copies of the attachments, name and actor doesn't matter here
object_ids_and_hrefs =
from(o in Object,
where:
fragment(
"to_jsonb(array(select jsonb_array_elements((?)#>'{url}') ->> 'href' where jsonb_typeof((?)#>'{url}') = 'array'))::jsonb \\?| (?)",
o.data,
o.data,
^hrefs
)
)
# The query above can be time consumptive on large instances until we
# refactor how uploads are stored
|> Repo.all(timeout: :infinity)
# we should delete 1 object for any given attachment, but don't delete
# files if there are more than 1 object for it
|> Enum.reduce(%{}, fn %{
id: id,
data: %{
"url" => [%{"href" => href}],
"actor" => obj_actor,
"name" => name
}
},
acc ->
Map.update(acc, href, %{id: id, count: 1}, fn val ->
case obj_actor == actor and name in names do
true ->
# set id of the actor's object that will be deleted
%{val | id: id, count: val.count + 1}
Enum.each(attachment_urls, fn href ->
href
|> String.trim_leading("#{base_url}/#{prefix}")
|> uploader.delete_file()
end)
false ->
# another actor's object, just increase count to not delete file
%{val | count: val.count + 1}
end
end)
end)
|> Enum.map(fn {href, %{id: id, count: count}} ->
# only delete files that have single instance
with 1 <- count do
href
|> String.trim_leading("#{base_url}/#{prefix}")
|> uploader.delete_file()
{id, href}
else
_ -> {id, nil}
end
end)
object_ids = Enum.map(object_ids_and_hrefs, fn {id, _} -> id end)
from(o in Object, where: o.id in ^object_ids)
|> Repo.delete_all()
object_ids_and_hrefs
|> Enum.filter(fn {_, href} -> not is_nil(href) end)
|> Enum.map(&elem(&1, 1))
|> Pleroma.Web.MediaProxy.Invalidation.purge()
{:ok, :success}
delete_objects(object_ids)
end
def perform(%{"op" => "cleanup_attachments", "object" => _object}, _job), do: {:ok, :skip}
defp delete_objects([_ | _] = object_ids) do
Repo.delete_all(from(o in Object, where: o.id in ^object_ids))
end
defp delete_objects(_), do: :ok
# we should delete 1 object for any given attachment, but don't delete
# files if there are more than 1 object for it
def prepare_objects(objects, actor, names) do
objects
|> Enum.reduce(%{}, fn %{
id: id,
data: %{
"url" => [%{"href" => href}],
"actor" => obj_actor,
"name" => name
}
},
acc ->
Map.update(acc, href, %{id: id, count: 1}, fn val ->
case obj_actor == actor and name in names do
true ->
# set id of the actor's object that will be deleted
%{val | id: id, count: val.count + 1}
false ->
# another actor's object, just increase count to not delete file
%{val | count: val.count + 1}
end
end)
end)
end
def fetch_objects(hrefs) do
from(o in Object,
where:
fragment(
"to_jsonb(array(select jsonb_array_elements((?)#>'{url}') ->> 'href' where jsonb_typeof((?)#>'{url}') = 'array'))::jsonb \\?| (?)",
o.data,
o.data,
^hrefs
)
)
# The query above can be time consumptive on large instances until we
# refactor how uploads are stored
|> Repo.all(timeout: :infinity)
end
end

View file

@ -0,0 +1,65 @@
defmodule Pleroma.Web.MediaProxy.InvalidationTest do
use ExUnit.Case
use Pleroma.Tests.Helpers
alias Pleroma.Config
alias Pleroma.Web.MediaProxy.Invalidation
import ExUnit.CaptureLog
import Mock
import Tesla.Mock
setup do: clear_config([:media_proxy])
setup do
on_exit(fn -> Cachex.clear(:deleted_urls_cache) end)
:ok
end
describe "Invalidation.Http" do
test "perform request to clear cache" do
Config.put([:media_proxy, :enabled], false)
Config.put([:media_proxy, :invalidation, :enabled], true)
Config.put([:media_proxy, :invalidation, :provider], Invalidation.Http)
Config.put([Invalidation.Http], method: :purge, headers: [{"x-refresh", 1}])
image_url = "http://example.com/media/example.jpg"
Pleroma.Web.MediaProxy.put_in_deleted_urls(image_url)
mock(fn
%{
method: :purge,
url: "http://example.com/media/example.jpg",
headers: [{"x-refresh", 1}]
} ->
%Tesla.Env{status: 200}
end)
assert capture_log(fn ->
assert Pleroma.Web.MediaProxy.in_deleted_urls(image_url)
assert Invalidation.purge([image_url]) == {:ok, [image_url]}
assert Pleroma.Web.MediaProxy.in_deleted_urls(image_url)
end) =~ "Running cache purge: [\"#{image_url}\"]"
end
end
describe "Invalidation.Script" do
test "run script to clear cache" do
Config.put([:media_proxy, :enabled], false)
Config.put([:media_proxy, :invalidation, :enabled], true)
Config.put([:media_proxy, :invalidation, :provider], Invalidation.Script)
Config.put([Invalidation.Script], script_path: "purge-nginx")
image_url = "http://example.com/media/example.jpg"
Pleroma.Web.MediaProxy.put_in_deleted_urls(image_url)
with_mocks [{System, [], [cmd: fn _, _ -> {"ok", 0} end]}] do
assert capture_log(fn ->
assert Pleroma.Web.MediaProxy.in_deleted_urls(image_url)
assert Invalidation.purge([image_url]) == {:ok, [image_url]}
assert Pleroma.Web.MediaProxy.in_deleted_urls(image_url)
end) =~ "Running cache purge: [\"#{image_url}\"]"
end
end
end
end

View file

@ -5,6 +5,11 @@ defmodule Pleroma.Web.MediaProxy.Invalidation.HttpTest do
import ExUnit.CaptureLog
import Tesla.Mock
setup do
on_exit(fn -> Cachex.clear(:deleted_urls_cache) end)
:ok
end
test "logs hasn't error message when request is valid" do
mock(fn
%{method: :purge, url: "http://example.com/media/example.jpg"} ->
@ -14,8 +19,8 @@ test "logs hasn't error message when request is valid" do
refute capture_log(fn ->
assert Invalidation.Http.purge(
["http://example.com/media/example.jpg"],
%{}
) == {:ok, "success"}
[]
) == {:ok, ["http://example.com/media/example.jpg"]}
end) =~ "Error while cache purge"
end
@ -28,8 +33,8 @@ test "it write error message in logs when request invalid" do
assert capture_log(fn ->
assert Invalidation.Http.purge(
["http://example.com/media/example1.jpg"],
%{}
) == {:ok, "success"}
[]
) == {:ok, ["http://example.com/media/example1.jpg"]}
end) =~ "Error while cache purge: url - http://example.com/media/example1.jpg"
end
end

View file

@ -4,17 +4,24 @@ defmodule Pleroma.Web.MediaProxy.Invalidation.ScriptTest do
import ExUnit.CaptureLog
setup do
on_exit(fn -> Cachex.clear(:deleted_urls_cache) end)
:ok
end
test "it logger error when script not found" do
assert capture_log(fn ->
assert Invalidation.Script.purge(
["http://example.com/media/example.jpg"],
%{script_path: "./example"}
) == {:error, "\"%ErlangError{original: :enoent}\""}
end) =~ "Error while cache purge: \"%ErlangError{original: :enoent}\""
script_path: "./example"
) == {:error, "%ErlangError{original: :enoent}"}
end) =~ "Error while cache purge: %ErlangError{original: :enoent}"
assert Invalidation.Script.purge(
["http://example.com/media/example.jpg"],
%{}
) == {:error, "not found script path"}
capture_log(fn ->
assert Invalidation.Script.purge(
["http://example.com/media/example.jpg"],
[]
) == {:error, "\"not found script path\""}
end)
end
end

View file

@ -10,6 +10,11 @@ defmodule Pleroma.Web.MediaProxy.MediaProxyControllerTest do
setup do: clear_config(:media_proxy)
setup do: clear_config([Pleroma.Web.Endpoint, :secret_key_base])
setup do
on_exit(fn -> Cachex.clear(:deleted_urls_cache) end)
:ok
end
test "it returns 404 when MediaProxy disabled", %{conn: conn} do
Config.put([:media_proxy, :enabled], false)
@ -66,4 +71,16 @@ test "it performs ReverseProxy.call when signature valid", %{conn: conn} do
assert %Plug.Conn{status: :success} = get(conn, url)
end
end
test "it returns 404 when url contains in deleted_urls cache", %{conn: conn} do
Config.put([:media_proxy, :enabled], true)
Config.put([Pleroma.Web.Endpoint, :secret_key_base], "00000000000")
url = Pleroma.Web.MediaProxy.encode_url("https://google.fn/test.png")
Pleroma.Web.MediaProxy.put_in_deleted_urls("https://google.fn/test.png")
with_mock Pleroma.ReverseProxy,
call: fn _conn, _url, _opts -> %Plug.Conn{status: :success} end do
assert %Plug.Conn{status: 404, resp_body: "Not Found"} = get(conn, url)
end
end
end